Dump RPM headers

dump-rpm-headers.py

#! /usr/bin/env python

import struct
from ctypes import *
import io
import re

from enum import Enum
from sys import argv

"""
Linux Standard Base Core Specification, Generic Part - 25.2. Package File Format
https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/pkgformat.html
"""

class IndexTypeValue(Enum):
    NULL = 0
    CHAR = 1
    INT8 = 2
    INT16 = 3
    INT32 = 4
    INT64 = 5
    STRING = 6
    BIN = 7
    STRING_ARRAY = 8
    I18NSTRING = 9

    def ctype_size(self):
        return {
            self.NULL: 1,
            self.CHAR: 1,
            self.INT8: 1,
            self.INT16: 2,
            self.INT32: 4,
            self.INT64: 8,
            self.STRING: 1,
            self.BIN: 1,
            self.STRING_ARRAY: 1,
            self.I18NSTRING: 1,
        }[self]

    def format_char(self):
        format_chars = {
            self.INT8: 'C',
            self.INT16: 'H',
            self.INT32: 'I',
            self.INT64: 'L',
        }
        if self in format_chars:
            return format_chars[self]
        return None


class SignatureTagValue(Enum):
    RPMTAG_HEADERSIGNATURES = 62
    RPMTAG_HEADERIMMUTABLE = 63
    RPMTAG_HEADERI18NTABLE = 100

    RPMSIGTAG_SIZE = 1000
    RPMSIGTAG_PAYLOADSIZE = 1007
    RPMSIGTAG_SHA1 = 269
    RPMSIGTAG_MD5 = 1004
    RPMSIGTAG_DSA = 267
    RPMSIGTAG_RSA = 268
    RPMSIGTAG_PGP = 1002
    RPMSIGTAG_GPG = 1005

class HeaderTagValue(Enum):
    RPMTAG_HEADERSIGNATURES = 62
    RPMTAG_HEADERIMMUTABLE = 63
    RPMTAG_HEADERI18NTABLE = 100

    RPMTAG_NAME = 1000
    RPMTAG_VERSION = 1001
    RPMTAG_RELEASE = 1002
    RPMTAG_SUMMARY = 1004
    RPMTAG_DESCRIPTION = 1005
    RPMTAG_SIZE = 1009
    RPMTAG_DISTRIBUTION = 1010
    RPMTAG_VENDOR = 1011
    RPMTAG_LICENSE = 1014
    RPMTAG_PACKAGER = 1015
    RPMTAG_GROUP = 1016
    RPMTAG_URL = 1020
    RPMTAG_OS = 1021
    RPMTAG_ARCH = 1022
    RPMTAG_SOURCERPM = 1044
    RPMTAG_ARCHIVESIZE = 1046
    RPMTAG_RPMVERSION = 1064
    RPMTAG_COOKIE = 1094
    RPMTAG_DISTURL = 1123
    RPMTAG_PAYLOADFORMAT = 1124
    RPMTAG_PAYLOADCOMPRESSOR = 1125
    RPMTAG_PAYLOADFLAGS = 1126
    RPMTAG_PREIN = 1023
    RPMTAG_POSTIN = 1024
    RPMTAG_PREUN = 1025
    RPMTAG_POSTUN = 1026
    RPMTAG_PREINPROG = 1085
    RPMTAG_POSTINPROG = 1086
    RPMTAG_PREUNPROG = 1087
    RPMTAG_POSTUNPROG = 1088
    RPMTAG_OLDFILENAMES = 1027
    RPMTAG_FILESIZES = 1028
    RPMTAG_FILEMODES = 1030
    RPMTAG_FILERDEVS = 1033
    RPMTAG_FILEMTIMES = 1034
    RPMTAG_FILEMD5S = 1035
    RPMTAG_FILELINKTOS = 1036
    RPMTAG_FILEFLAGS = 1037
    RPMTAG_FILEUSERNAME = 1039
    RPMTAG_FILEGROUPNAME = 1040
    RPMTAG_FILEDEVICES = 1095
    RPMTAG_FILEINODES = 1096
    RPMTAG_FILELANGS = 1097
    RPMTAG_DIRINDEXES = 1116
    RPMTAG_BASENAMES = 1117
    RPMTAG_DIRNAMES = 1118
    RPMTAG_PROVIDENAME = 1047
    RPMTAG_REQUIREFLAGS = 1048
    RPMTAG_REQUIRENAME = 1049
    RPMTAG_REQUIREVERSION = 1050
    RPMTAG_CONFLICTFLAGS = 1053
    RPMTAG_CONFLICTNAME = 1054
    RPMTAG_CONFLICTVERSION = 1055
    RPMTAG_OBSOLETENAME = 1090
    RPMTAG_PROVIDEFLAGS = 1112
    RPMTAG_PROVIDEVERSION = 1113
    RPMTAG_OBSOLETEFLAGS = 1114
    RPMTAG_OBSOLETEVERSION = 1115
    RPMTAG_BUILDTIME = 1006
    RPMTAG_BUILDHOST = 1007
    RPMTAG_FILEVERIFYFLAGS = 1045
    RPMTAG_CHANGELOGTIME = 1080
    RPMTAG_CHANGELOGNAME = 1081
    RPMTAG_CHANGELOGTEXT = 1082
    RPMTAG_OPTFLAGS = 1122
    RPMTAG_RHNPLATFORM = 1131
    RPMTAG_PLATFORM = 1132

    # https://github.com/rpm-software-management/rpm/blob/847c6f062c267c4be643be9c202141bd330a7891/lib/rpmtag.h
    RPMTAG_PATCHESNAME          = 1133
    RPMTAG_PATCHESFLAGS         = 1134
    RPMTAG_PATCHESVERSION       = 1135
    RPMTAG_CACHECTIME           = 1136
    RPMTAG_CACHEPKGPATH         = 1137
    RPMTAG_CACHEPKGSIZE         = 1138
    RPMTAG_CACHEPKGMTIME        = 1139
    RPMTAG_FILECOLORS           = 1140
    RPMTAG_FILECLASS            = 1141
    RPMTAG_CLASSDICT            = 1142
    RPMTAG_FILEDEPENDSX         = 1143
    RPMTAG_FILEDEPENDSN         = 1144
    RPMTAG_DEPENDSDICT          = 1145
    RPMTAG_SOURCEPKGID          = 1146
    RPMTAG_FILECONTEXTS         = 1147
    RPMTAG_FSCONTEXTS           = 1148
    RPMTAG_RECONTEXTS           = 1149
    RPMTAG_POLICIES             = 1150
    RPMTAG_PRETRANS             = 1151
    RPMTAG_POSTTRANS            = 1152
    RPMTAG_PRETRANSPROG         = 1153
    RPMTAG_POSTTRANSPROG        = 1154
    RPMTAG_DISTTAG              = 1155
    RPMTAG_OLDSUGGESTSNAME      = 1156

    RPMTAG_FILEDIGESTALGO = 5011
    RPMTAG_BUGURL = 5012


class RPMLead(BigEndianStructure):
    _fields_ = (
        ('magic', c_char * 4),
        ('major', c_char),
        ('minor', c_char),
        ('type', c_short),
        ('archnum', c_short),
        ('name', c_char * 66),
        ('osnum', c_short),
        ('signature_type', c_short),
        ('reserved', c_char * 16),
    )

class RPMHeader(BigEndianStructure):
    _fields_ = (
        ('magic', c_char * 3),
        ('version', c_char),
        ('reserved', c_char * 4),
        ('nindex', c_int),
        ('hsize', c_int),
    )

class RPMHeaderIndex(BigEndianStructure):
    position = -1
    tag_name = "Unknown"
    _fields_ = (
        ('tag', c_int),
        ('type', c_int),
        ('offset', c_int),
        ('count', c_int),
    )

def read_string(source:io.IOBase):
    result = []
    while True:
        chunk = source.read(128)
        i = chunk.find(b'\0')
        if i == -1:
            result.append(chunk)
        else:
            result.append(chunk[:i])
            source.seek(i - 128 + 1, io.SEEK_CUR)
            break
        if len(chunk) < 128:
            break
    return b''.join(result)

def read_string_array(source:io.IOBase, length:int):
    result = []
    for i in range(length):
        result.append(read_string(source))
    return result

def main(rpmfile):
    source = open(rpmfile, 'rb')

    print("==== Lead Section ====")

    lead = RPMLead()
    source.readinto(lead)

    for field in lead._fields_:
        v = getattr(lead, field[0])
        print(f"{field[0]:>10}: {v}")

    dump_header_structure(source, "Signature Section", SignatureTagValue)
    dump_header_structure(source, "Header Section", HeaderTagValue)

    source.close()

def dump_header_structure(source, label, tag_value):
    print(f"==== {label} - Header ====")

    header = RPMHeader()
    source.readinto(header)

    for field in header._fields_:
        v = getattr(header, field[0])
        print(f"{field[0]:>10}: {v}")

    print(f"==== {label} - Index Entries ====")

    index_entries = []
    for i in range(header.nindex):
        e = RPMHeaderIndex()
        source.readinto(e)
        e.position = i
        index_entries.append(e)

        try:
            e.tag_name = tag_value(e.tag).name
        except:
            pass

        t = IndexTypeValue(e.type).name
        print(f"[{i:>3}] tag:{e.tag:>5}  type: {t:<12} offset:{e.offset:>4}  count:{e.count:>4}   # {e.tag_name}")

    print(f"==== {label} - Data ====")

    index_entries = sorted(index_entries, key=lambda e: e.offset)

    start = source.tell()
    for index in index_entries:
        index_type = IndexTypeValue(index.type)
        size = index_type.ctype_size()

        source.seek(start + index.offset)

        if index_type == IndexTypeValue.STRING:
            v = read_string(source)
        elif index_type == IndexTypeValue.STRING_ARRAY:
            v = read_string_array(source, index.count)
            v = '  ' + '\n  '.join(map(str, v))
        else:
            v = source.read(size * index.count)
            format_char = index_type.format_char()
            if format_char:
                v = '  ' + '\n  '.join(map(str, struct.unpack('>' + format_char * index.count, v)))
            else:
                v = re.sub('([0-9a-f]{4})', '\\1 ', re.sub('(.{32})', '\\1\n', v.hex()))
                v = v.strip()
        print(f"[{index.position:>3}] {index.tag_name}\n{v}")
        print("")

main(argv[1])